frankcrc commented on issue #5174:
URL: https://github.com/apache/rocketmq/issues/5174#issuecomment-1260402596

   Thanks for reply, @ni-ze @ShadowySpirits .
   
   From the code, this problem could reproduce as long as the `storeTimestamp` 
less than `bornTimestamp`, even if `storeTimestamp` were just 1ms behind.
   
   I wanna know if there a recommanded time difference threshold between client 
and broker. Let say 20ms, if the time difference were larger than this, this 
problem would be more likely to happen.
   
   > IMO, this is not a bug. Most APIs in the broker, such as 
SEARCH_OFFSET_BY_TIMESTAMP, and QUERY_MESSAGE, use store timestamp, not born 
timestamp. Because multiple clients are more likely has clock synchronization 
problems.
   
   If client can specify timestamp, that's ok. But all the timestamps used in 
`consumeMessageDirectly` are calculated inner, client cannot control unless 
implements a special `MqAdminExt`.
   
   So I think , maybe, we could modify 'consumeMessageDirectly',  add a 
optional timestamp parameter to it.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to